Here's the tty and serial driver patches for 4.2-rc1.
A number of individual driver updates, some code cleanups, and other
minor things, full details in the shortlog.
All have been in linux-next for a while with no reported issues.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
iEYEABECAAYFAlWNoSAACgkQMUfUDdst+ymxNQCguSEmkAYNDdLyYhdcOqSxJt9u
U1gAoMThUDoomkx6CTDMU1wn53hxgMk9
=eCUS
-----END PGP SIGNATURE-----
Merge tag 'tty-4.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Pull tty/serial driver updates from Greg KH:
"Here's the tty and serial driver patches for 4.2-rc1.
A number of individual driver updates, some code cleanups, and other
minor things, full details in the shortlog.
All have been in linux-next for a while with no reported issues"
* tag 'tty-4.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty: (152 commits)
Doc: serial-rs485.txt: update RS485 driver interface
Doc: tty.txt: remove mention of the BKL
MAINTAINERS: tty: add serial docs directory
serial: sprd: check for NULL after calling devm_clk_get
serial: 8250_pci: Correct uartclk for xr17v35x expansion chips
serial: 8250_pci: Add support for 12 port Exar boards
serial: 8250_uniphier: add bindings document for UniPhier UART
serial: core: cleanup in uart_get_baud_rate()
serial: stm32-usart: Add STM32 USART Driver
tty/serial: kill off set_irq_flags usage
tty: move linux/gsmmux.h to uapi
doc: dt: add documentation for nxp,lpc1850-uart
serial: 8250: add LPC18xx/43xx UART driver
serial: 8250_uniphier: add UniPhier serial driver
serial: 8250_dw: support ACPI platforms with integrated DMA engine
serial: of_serial: check the return value of clk_prepare_enable()
serial: of_serial: use devm_clk_get() instead of clk_get()
serial: earlycon: Add support for big-endian MMIO accesses
serial: sirf: use hrtimer for data rx
serial: sirf: correct the fifo empty_bit
...
Since 39b2bbe3d7 (gpio: add flags argument to gpiod_get*() functions),
the gpiod_get* functions take an additional parameter that allows to
specify direction and initial value for output.
Simplify the usage of devm_gpiod_get_optional accordingly.
Signed-off-by: Axel Lin <axel.lin@ingics.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
Don't clear the backlight level when we're going into suspend or
blanking. Instead, just temporarily set the level to 0 so we retain
the value when we resume.
Reported-by: Benson Leung <bleung@chromium.org>
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Tested-by: Stephen Barber <smbarber@chromium.org>
Reviewed-by: Benson Leung <bleung@chromium.org>
Acked-by: Milo Kim <milo.kim@ti.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
The GPIO subsystem provides dummy GPIO consumer functions if GPIOLIB is
not enabled. Hence drivers that depend on GPIOLIB, but use GPIO consumer
functionality only, can still be compiled if GPIOLIB is not enabled.
Relax the dependency on GPIOLIB if COMPILE_TEST is enabled, where
appropriate.
Cc: Jingoo Han <jg1.han@samsung.com>
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
The platform_device_id is not modified by the driver and core uses it as
const.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski.k@gmail.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
The gpio-backlight driver seems to be missing the
MODULE_DEVICE_TABLE line which is preventing it from
being modprobed during boot time even if the
gpio-backlight device exists.
This seems to be a bug and this patch attempts to
fix that.
Signed-off-by: Arun Bharadwaj <arun@gumstix.com>
Signed-off-by: Ash Charles <ashcharles@gmail.com>
Signed-off-by: Lee Jones <lee.jones@linaro.org>
This code is no longer used now that mach-msm has been removed.
Delete it.
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: David Brown <davidb@codeaurora.org>
Cc: Bryan Huntsman <bryanh@codeaurora.org>
Cc: Daniel Walker <dwalker@fifo99.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj@jcrosoft.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
At the moment when HDMI video output is stopped, we just clear the
enable bit and return. While it's unclear if this can cause any issues,
I think it's still better to wait for FRAMEDONE interrupt after clearing
the enable bit so that we're sure the HDMI IP has finished.
As we don't have any ready-made irq handling for HDMI, and this only
needs to be done when disabling the HDMI output, this patch implements a
simple loop with sleep, polling the FRAMEDONE bit.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Error handling in hdmi_power_on_full() is not correct, and could leave
resources unfreed.
Fix this by arranging the error labels correctly.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The current driver does both x and y decimation on OMAP3 DSS. Testing
shows that x decimation rarely works, leading to underflows.
The exact reason for this is unclear, as the underflows seem to happen
even with low pixel clock rates, and I would presume that if the DSS can
manage a display with 140MHz pixel clock, it could manage x decimation
with factor 2 with a low pixel clock (~30MHz).
So it is possible that there is a problem somewhere else, in memory
management, or DSS DMA, or similar. I have not found anything that would
help this.
So, to fix the downscaling scaling, this patch removes x decimation for
OMAP3. This will limit some of the more demanding downscaling scenarios,
but one could argue that using DSS to downscale such a large amount is
insane in the first place, as the produced image is rather bad quality.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The DISPC's scaling code seems to presume that decimation always
succeeds, and so we always do find a suitable downscaling setup.
However, this is not the case, and the algorithm can fail.
When that happens, the code just proceeds with wrong results, causing
issues later.
Add the necessary checks to bail out if the scaling algorithm failed.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The DISPC driver uses 64 bit arithmetic to calculate the required clock
rate for scaling. The code does not seem to work correctly, and instead
calculates with 32 bit numbers, giving wrong result.
Fix the code by typecasting values to u64 first, so that the
calculations do happen in 64 bits.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
pixel_inc and row_inc work differently on OMAP2/3 and OMAP4+ DSS. On
OMAP2/3 DSS, the pixel_inc is _not_ added by the HW at the end of the
line, after the last pixel, whereas on OMAP4+ it is.
The driver currently works for OMAP4+, but does not handle OMAP2/3
correctly, which leads to tilted image when row_inc is used.
This patch adds a flag to DISPC driver so that the pixel_inc is added
when required.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
On OMAP3/AM43xx some scaling factors cause underflows/synclosts. After
studying this, I found that sometimes the driver uses three-tap scaling
with downscaling factor smaller than x0.5. This causes issues, as x0.5
is the limit for three-tap scaling.
The driver has FEAT_PARAM_DOWNSCALE parameter, but that seems to be for
five-tap scaling, which allows scaling down to x0.25.
This patch adds checks for both horizontal and vertical scaling. For
horizontal the HW always uses 5 taps, so the limit is x0.25.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
After calculating the required decimation for scaling, the dispc driver
checks once more if the resulting configuration is valid by calling
check_horiz_timing_omap3().
Earlier calls to this function have correctly used in_width and
in_height as parameters, but the last call uses width and height. This
causes the driver to possibly reject scaling that would work.
This patch fixes the parameters.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
DISPC needs even input buffer width for YUV modes. The DISPC driver
doesn't check this at the moment (although omapdrm does), but worse,
when DISPC driver does x predecimation the result may be uneven. This
causes sometimes sync losts, underflows, or just visual errors.
This patch makes DISPC driver return an error if the user gives uneven
input width for a YUV buffer. It also makes the input width even in case
of predecimation.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Errata i631 description:
"When in YUV4:2:0 format in 1D burst, the DISPC DMA skips lines when
fetching Chroma sampling."
Workaround:
"If YUV4:2:0-1D burst is required: Set
DISPC_VIDp_ATTRIBUTES[22]DOUBLESTRIDE to 0x0 and
DISPC_VIDp_ATTRIBUTES[13:12]ROTATION to 0x1 or 0x3"
The description is somewhat confusing, but testing has shown that DSS
fetches extra rows from memory when using NV12 format in 1D mode. If the
memory after the framebuffer is inaccessible, this leads to OCP errors.
The driver always uses DOUBLESTRIDE=0 when using 1D mode, so we only
need to handle the ROTATION part.
The issue exist on all OMAP4 and OMAP5 based DSS IPs.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Now that we are using components in omapdss, there's no need for
separate handling of dss and dispc driver init. Thus we can move the dss
and dispc init and unit func pointers to the lists we use for the other
dss submodules.
We can now also handle errors returned by the registration functions
properly: if registering a driver fails, we can stop processing and
return the error.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
omapdss kernel module contains drivers for multiple devices, one for
each DSS submodule. The probing we have at the moment is a mess, and
doesn't give us proper deferred probing nor ensure that all the devices
are probed before omapfb/omapdrm start using omapdss.
This patch solves the mess by using the component system for DSS
submodules.
The changes to all DSS submodules (dispc, dpi, dsi, hdmi4/5, rfbi, sdi,
venc) are the same: probe & remove functions are changed to bind &
unbind, and new probe & remove functions are added which call
component_add/del.
The dss_core driver (dss.c) acts as a component master. Adding and
matching the components is simple: all dss device's child devices are
added as components.
However, we do have some dependencies between the drivers. The order in
which they should be probed is reflected by the list in core.c
(dss_output_drv_reg_funcs). The drivers are registered in that order,
which causes the components to be added in that order, which makes the
components to be bound in that order. This feels a bit fragile, and we
probably should improve the code to manage binds in random order.
However, for now, this works fine.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
We have a list of function pointers to dss submodule uninit functions.
It makes sense to do the uninit in the reverse order to init, but that
is not currently the case.
This patch reorders the uninit calls to be the reverse of init order.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The following patches will add component handling to omapdss, improving
the handling of deferred probing. However, at the moment we're using
quite a lot of __inits and __exits in the driver, which prevent normal
dynamic probing and removal.
This patch removes most of the uses of __init and __exit, so that we can
register drivers after module init, and so that we can unregister
drivers even if the module is built-in.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
The return value of dss_init_ports() is not handled at all, causing
crashes later if the call failed.
This patch adds the error handling, and we also move the call to a
slightly earlier place to make bailing out easier.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Refactor dss probe function by extracting the setup for video plls into
a separate function. The call to this function is also moved to a
slightly earlier phase, so that in error case we can bail out more
easily.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
We have a flag, 'dss_initialized', which tells omapfb and omapdrm if
omapdss is available. At the moment it can be set even if the dss
submodules are not all ready, in case something gets deferred.
Move the flag to dss_core driver so that it'll signal the availability
of the dss drivers move accurately.
For now, it'll signal that dss_core is ready, which is not quite correct
but still better than previously. The following patches will add
component system to omapdss, and after those patches 'dss_initialized'
will signal that all the submodules are ready.
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
fb_videomode_from_videomode() may fail, but of_get_fb_videomode()
silently covers this fact. Instead, trow the error code to the
caller.
Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
This driver uses the same area for MTRR as for the ioremap_wc(), if
anything it just uses a smaller size in case MTRR reservation fails.
ioremap_wc() API is already used to take advantage of architecture
write-combining when available.
Convert the driver from using the x86 specific MTRR code to
the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add()
will avoid MTRR if write-combining is available.
There are a few motivations for this:
a) Take advantage of PAT when available
b) Help bury MTRR code away, MTRR is architecture specific and on
x86 its replaced by PAT
c) Help with the goal of eventually using _PAGE_CACHE_UC over
_PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit
de33c442e titled "x86 PAT: fix performance drop for glx,
use UC minus for ioremap(), ioremap_nocache() and
pci_mmap_page_range()")
The conversion done is expressed by the following Coccinelle
SmPL patch, it additionally required manual intervention to
address all the #ifdery and removal of redundant things which
arch_phys_wc_add() already addresses such as verbose message
about when MTRR fails and doing nothing when we didn't get
an MTRR.
@ mtrr_found @
expression index, base, size;
@@
-index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1);
+index = arch_phys_wc_add(base, size);
@ mtrr_rm depends on mtrr_found @
expression mtrr_found.index, mtrr_found.base, mtrr_found.size;
@@
-mtrr_del(index, base, size);
+arch_phys_wc_del(index);
@ mtrr_rm_zero_arg depends on mtrr_found @
expression mtrr_found.index;
@@
-mtrr_del(index, 0, 0);
+arch_phys_wc_del(index);
@ mtrr_rm_fb_info depends on mtrr_found @
struct fb_info *info;
expression mtrr_found.index;
@@
-mtrr_del(index, info->fix.smem_start, info->fix.smem_len);
+arch_phys_wc_del(index);
@ ioremap_replace_nocache depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap_nocache(base, size);
+info->screen_base = ioremap_wc(base, size);
@ ioremap_replace_default depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap(base, size);
+info->screen_base = ioremap_wc(base, size);
Generated-by: Coccinelle SmPL
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Jingoo Han <jg1.han@samsung.com>
Cc: Wolfram Sang <wsa@the-dreams.de>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The MTRR added was never being deleted, in order to store the
MTRR cookie we need to make use of the private info->par so we
create a struct for this. This driver was already using the extra
space typically used for info->par for the info->pseudo_palette
which typically used stuffed on driver's own private structs
(the respective info->par), so we just move the pseudo_palette
into the private struct as well.
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Jingoo Han <jg1.han@samsung.com>
Cc: Wolfram Sang <wsa@the-dreams.de>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
No other video driver uses MTRR types except for MTRR_TYPE_WRCOMB,
the other MTRR types were implemented and supported here but with
no real good reason. The ioremap() APIs are architecture agnostic and
at least on x86 PAT is a new design that extends MTRRs and
can replace it in a much cleaner way, where so long as the
proper ioremap_wc() or variant API is used the right thing will
be done behind the scenes. This is the only driver left using the
other MTRR types -- and since there is no good reason for it now
rip them out.
Cc: Toshi Kani <toshi.kani@hp.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Jingoo Han <jg1.han@samsung.com>
Cc: Wolfram Sang <wsa@the-dreams.de>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The platform_device_id is not modified by the driver and core uses it as
const.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski.k@gmail.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The platform_device_id is not modified by the driver and core uses it as
const.
Signed-off-by: Krzysztof Kozlowski <k.kozlowski.k@gmail.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Since kobject_init_and_add takes a format string, make sure that the
passed in name cannot be accidentally parsed.
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
This fixes up a merge issue with the amba-pl011.c driver, and we want
the fixes in this branch as well.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There is a constraint in the OMAP4 HDMI IP that requires to use
the 8-channel code when transmitting more than two channels.
The constraint doesn't apply for OMAP5 so don't force the channel
allocation in the sound driver as it can be done specifically for
OMAP4 later in the hdmi4 core.
Signed-off-by: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Acked-by: Mark Brown <broonie@kernel.org>
Downmix inhibit in HDMI_CORE_FC_AUDICONF3 register is in
bit 4 while CEA861_AUDIO_INFOFRAME_DB5_DM_INH sets bit 7.
Signed-off-by: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
As per TRM, HDMI_WP_AUDIO_CFG[2] LEFT_BEFORE = 0 is reserved,
so it must always be set to 1 (the first sample is the left).
Signed-off-by: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
OMAP4 HDMI IP uses the 8-channel layout with 8-channel speaker
allocation mask when transmitting more than two channels. But
the channel count field (CC) of the Audio InfoFrame's DB1 is
not updated for 8-channels.
As per HDMI Compliance Test 7.31 "Audio InfoFrame", CC = 7 is
required for 8-channels CA masks (0x13 and 0x1F).
Signed-off-by: Misael Lopez Cruz <misael.lopez@ti.com>
Signed-off-by: Jyri Sarha <jsarha@ti.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Since 39b2bbe3d7 (gpio: add flags argument to gpiod_get*() functions)
which appeared in v3.17-rc1, the gpiod_get* functions take an additional
parameter that allows to specify direction and initial value for output.
Also make use of gpiod_get*_optional where applicable.
Apart from simplification of the affected drivers this is another step
towards making the flags argument to gpiod_get*() mandatory.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The driver doesn't use mtrr_add() or arch_phys_wc_add() but
since we know the framebuffer is isolated already on an
ioremap() we can take advantage of write combining for
performance where possible.
In this case there are a few motivations for this:
a) Take advantage of PAT when available
b) Help with the goal of eventually using _PAGE_CACHE_UC over
_PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit
de33c442e titled "x86 PAT: fix performance drop for glx,
use UC minus for ioremap(), ioremap_nocache() and
pci_mmap_page_range()")
Cc: Sudip Mukherjee <sudipm.mukherjee@gmail.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Jingoo Han <jg1.han@samsung.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-geode@lists.infradead.org
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
The driver doesn't use mtrr_add() or arch_phys_wc_add() but
since we know the framebuffer is isolated already on an
ioremap() we can take advantage of write combining for
performance where possible.
In this case there are a few motivations for this:
a) Take advantage of PAT when available
b) Help with the goal of eventually using _PAGE_CACHE_UC over
_PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit
de33c442e titled "x86 PAT: fix performance drop for glx,
use UC minus for ioremap(), ioremap_nocache() and
pci_mmap_page_range()")
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
This driver uses the same area for MTRR as for the ioremap().
Convert the driver from using the x86 specific MTRR code to
the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add()
will avoid MTRR if write-combining is available, in order to
take advantage of that also ensure the ioremap'd area is requested
as write-combining.
There are a few motivations for this:
a) Take advantage of PAT when available
b) Help bury MTRR code away, MTRR is architecture specific and on
x86 its replaced by PAT
c) Help with the goal of eventually using _PAGE_CACHE_UC over
_PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit
de33c442e titled "x86 PAT: fix performance drop for glx,
use UC minus for ioremap(), ioremap_nocache() and
pci_mmap_page_range()")
The conversion done is expressed by the following Coccinelle
SmPL patch, it additionally required manual intervention to
address all the #ifdery and removal of redundant things which
arch_phys_wc_add() already addresses such as verbose message
about when MTRR fails and doing nothing when we didn't get
an MTRR.
@ mtrr_found @
expression index, base, size;
@@
-index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1);
+index = arch_phys_wc_add(base, size);
@ mtrr_rm depends on mtrr_found @
expression mtrr_found.index, mtrr_found.base, mtrr_found.size;
@@
-mtrr_del(index, base, size);
+arch_phys_wc_del(index);
@ mtrr_rm_zero_arg depends on mtrr_found @
expression mtrr_found.index;
@@
-mtrr_del(index, 0, 0);
+arch_phys_wc_del(index);
@ mtrr_rm_fb_info depends on mtrr_found @
struct fb_info *info;
expression mtrr_found.index;
@@
-mtrr_del(index, info->fix.smem_start, info->fix.smem_len);
+arch_phys_wc_del(index);
@ ioremap_replace_nocache depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap_nocache(base, size);
+info->screen_base = ioremap_wc(base, size);
@ ioremap_replace_default depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap(base, size);
+info->screen_base = ioremap_wc(base, size);
Generated-by: Coccinelle SmPL
Cc: Rob Clark <robdclark@gmail.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
This driver uses the same area for MTRR as for the ioremap().
Convert the driver from using the x86 specific MTRR code to
the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add()
will avoid MTRR if write-combining is available, in order to
take advantage of that also ensure the ioremap'd area is requested
as write-combining.
There are a few motivations for this:
a) Take advantage of PAT when available
b) Help bury MTRR code away, MTRR is architecture specific and on
x86 its replaced by PAT
c) Help with the goal of eventually using _PAGE_CACHE_UC over
_PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit
de33c442e titled "x86 PAT: fix performance drop for glx,
use UC minus for ioremap(), ioremap_nocache() and
pci_mmap_page_range()")
The conversion done is expressed by the following Coccinelle
SmPL patch, it additionally required manual intervention to
address all the #ifdery and removal of redundant things which
arch_phys_wc_add() already addresses such as verbose message
about when MTRR fails and doing nothing when we didn't get
an MTRR.
@ mtrr_found @
expression index, base, size;
@@
-index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1);
+index = arch_phys_wc_add(base, size);
@ mtrr_rm depends on mtrr_found @
expression mtrr_found.index, mtrr_found.base, mtrr_found.size;
@@
-mtrr_del(index, base, size);
+arch_phys_wc_del(index);
@ mtrr_rm_zero_arg depends on mtrr_found @
expression mtrr_found.index;
@@
-mtrr_del(index, 0, 0);
+arch_phys_wc_del(index);
@ mtrr_rm_fb_info depends on mtrr_found @
struct fb_info *info;
expression mtrr_found.index;
@@
-mtrr_del(index, info->fix.smem_start, info->fix.smem_len);
+arch_phys_wc_del(index);
@ ioremap_replace_nocache depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap_nocache(base, size);
+info->screen_base = ioremap_wc(base, size);
@ ioremap_replace_default depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap(base, size);
+info->screen_base = ioremap_wc(base, size);
Generated-by: Coccinelle SmPL
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
This driver uses the same area for MTRR as for the ioremap().
Convert the driver from using the x86 specific MTRR code to
the architecture agnostic arch_phys_wc_add(). arch_phys_wc_add()
will avoid MTRR if write-combining is available, in order to
take advantage of that also ensure the ioremap'd area is requested
as write-combining.
There are a few motivations for this:
a) Take advantage of PAT when available
b) Help bury MTRR code away, MTRR is architecture specific and on
x86 its replaced by PAT
c) Help with the goal of eventually using _PAGE_CACHE_UC over
_PAGE_CACHE_UC_MINUS on x86 on ioremap_nocache() (see commit
de33c442e titled "x86 PAT: fix performance drop for glx,
use UC minus for ioremap(), ioremap_nocache() and
pci_mmap_page_range()")
The conversion done is expressed by the following Coccinelle
SmPL patch, it additionally required manual intervention to
address all the #ifdery and removal of redundant things which
arch_phys_wc_add() already addresses such as verbose message
about when MTRR fails and doing nothing when we didn't get
an MTRR.
@ mtrr_found @
expression index, base, size;
@@
-index = mtrr_add(base, size, MTRR_TYPE_WRCOMB, 1);
+index = arch_phys_wc_add(base, size);
@ mtrr_rm depends on mtrr_found @
expression mtrr_found.index, mtrr_found.base, mtrr_found.size;
@@
-mtrr_del(index, base, size);
+arch_phys_wc_del(index);
@ mtrr_rm_zero_arg depends on mtrr_found @
expression mtrr_found.index;
@@
-mtrr_del(index, 0, 0);
+arch_phys_wc_del(index);
@ mtrr_rm_fb_info depends on mtrr_found @
struct fb_info *info;
expression mtrr_found.index;
@@
-mtrr_del(index, info->fix.smem_start, info->fix.smem_len);
+arch_phys_wc_del(index);
@ ioremap_replace_nocache depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap_nocache(base, size);
+info->screen_base = ioremap_wc(base, size);
@ ioremap_replace_default depends on mtrr_found @
struct fb_info *info;
expression base, size;
@@
-info->screen_base = ioremap(base, size);
+info->screen_base = ioremap_wc(base, size);
Generated-by: Coccinelle SmPL
Cc: Jingoo Han <jg1.han@samsung.com>
Cc: Dan Carpenter <dan.carpenter@oracle.com>
Cc: Rob Clark <robdclark@gmail.com>
Cc: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Cc: Suresh Siddha <sbsiddha@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Juergen Gross <jgross@suse.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Antonino Daplas <adaplas@gmail.com>
Cc: Jean-Christophe Plagniol-Villard <plagnioj@jcrosoft.com>
Cc: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: linux-fbdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
